Skip to main content
Home/Challenges/Analysis/Entity Extractor
BeginnerAnalysis~5 min

Entity Extractor

Extract all named entities from a news headline — categorised by type — with zero hallucinations.

Skills Tested

Named entity recognitionJSON outputSchema adherence

Challenge Prompt

This is the example prompt structure. Write your own variation below.

Extract all named entities from this headline: "Apple and Samsung agree £2.4bn licensing deal as EU competition chief visits Seoul." Output as a JSON object with keys: PERSON, ORGANISATION, LOCATION, MONETARY_VALUE, EVENT. Each key maps to an array of strings. If none found for a type, output an empty array. Output valid JSON only — no prose.

Your Prompt

Write your own version. Every word matters — be specific.

🪙
Reward
+50 GeraCoins

Earn GeraCoins for completing this challenge. Redeem across all Gera products.

Hints (3)
  • 1"Valid JSON only — no prose" eliminates explanatory text that breaks JSON parsers.
  • 2Specifying all 5 entity types prevents the AI from inventing new categories.
  • 3Empty arrays for unfound types create a consistent schema downstream.

Related Topics

entity extractionNERJSON outputnews headlinestructured data
Browse All Challenges